[bigshot]

You professional consensus of one, i.e. you is not enough for me, considering how narrow your opinions have been in the past.

I find Gregorio to be very well informed. Even when I disagree with his opinions, it's only because he's applying the principles for different purposes than I am, not because his opinion is off base.

I'm sure that when transducers are manufactured, they need to settle into the groove and moving them around a little will do that. But I can't see that taking more than a minute or two, and I'm sure that they test things at the factory before shipping them out and accomplish that. Burn in of headphones sounds to me like the idea that silver wires sound brighter... it's just an analogy that we wallpaper over the truth because it seems correct. The reason that burn in with headphones hasn't been proven or disproven is because it is hard to accurately measure headphones without variation because of mechanical issues, and it's extremely difficult (if not impossible) to conduct blind line level matched direct A/B switched listening tests. On the other hand, we know for a fact that human hearing does acclimate and an unfamiliar response will sound better over time. The fact that almost everyone sees burn in as an improvement in sound over time, and no one ever seems to see it as getting worse, indicates to me that acclimating ears is the reason people report perceiving burn in.

 

[jagwap]

This is back to the circular arguement:

SS: We cannot measure it so it is not there.

Aph: But it does not visibly affect frequency response measurements.

SS: But we cannot measure other things, and frequency response indicates sound so it isn't there.

Aph: What about other measurements

SS: We don't have that ability. So it isn't there.

Aph: There are measurements published on the internet. We showed you them.

SS: If they didn't affect frequency response it isn't audable.

How about this: the measurement I linked to show a significant change on a woofer. Things like Fs changed by over 10%. That would affect low frequency behaviour the most. Low frequencies are extremely difficult to measure. Even well set up professional design houses have difficulty below a couple of hundred Hz, so someone in a studio or home has little to no chance of getting repeatable results.

Ok, so what about tweeters. If they are affected the same way we should be able to measure it? Not necessarily, as the change is likely below the crossover point, so not visible to frequency response measurements, unless you measure the un-crossed-over driver.

So it's not audible right? Not so. If you understand the mathmatical relationship between roll off in frequency response and phase and group delay you will know that the later two start becoming significant several octaves above thw roll off.

Also things like Fs change the Q of an acoustic system, causing the interaction of the driver and the cabinet and port to change. This affects the phase, group delay and damping of the system.

So this again is about speakers, and may or may not be valid in headphones, like some Schrodinger's cat of arguements. But do not ignore the evidence published because it doesn't fit your beliefs.

Frequency response is important. It is the macro performance criterion. But just because you cannot measure the other aspects, don't think they are trivial. They matter just as much in particular areas.

 

[castleofargh]

This is back to the circular arguement:

SS: We cannot measure it so it is not there.

Aph: But it does not visibly affect frequency response measurements.

SS: But we cannot measure other things, and frequency response indicates sound so it isn't there.

Aph: What about other measurements

SS: We don't have that ability. So it isn't there.

Aph: There are measurements published on the internet. We showed you them.

SS: If they didn't affect frequency response it isn't audable.

How about this: the measurement I linked to show a significant change on a woofer. Things like Fs changed by over 10%. That would affect low frequency behaviour the most. Low frequencies are extremely difficult to measure. Even well set up professional design houses have difficulty below a couple of hundred Hz, so someone in a studio or home has little to no chance of getting repeatable results.

Ok, so what about tweeters. If they are affected the same way we should be able to measure it? Not necessarily, as the change is likely below the crossover point, so not visible to frequency response measurements, unless you measure the un-crossed-over driver.

So it's not audible right? Not so. If you understand the mathmatical relationship between roll off in frequency response and phase and group delay you will know that the later two start becoming significant several octaves above thw roll off.

Also things like Fs change the Q of an acoustic system, causing the interaction of the driver and the cabinet and port to change. This affects the phase, group delay and damping of the system.

So this again is about speakers, and may or may not be valid in headphones, like some Schrodinger's cat of arguements. But do not ignore the evidence published because it doesn't fit your beliefs.

Frequency response is important. It is the macro performance criterion. But just because you cannot measure the other aspects, don't think they are trivial. They matter just as much in particular areas.

to me this would come down to 2 possibilities:
1/ there is a physical change in the drivers, but they don't end up affecting the outgoing signal in any significant way. so who cares?
2/ BS excuse. why wouldn't we be able to measure clearly audible change from the same source with the same mic over time? that's something I really don't get. if there is something I'm missing you guys need to explain it to me once and for all. characterizing a change could be real hard. identifying one variable as sole cause of a subjective impressions is often impossible. identifying a cause would probably require to observe the driver directly with lasers or other optical solution at really high speed and accuracy instead of looking at the sound. but if something is changing audibly over time, then it means the air pressure is changing in magnitude, else how do we hear it? and if a change is as I expect as simple as a change in the amplitudes of a signal over time, why should it be hard to record and confirm it, even with something as dumb as a null between 2 recordings at different time?
I just can't wrap my head around this. I have limitation because my rig is super cheap and my room not so quiet. but I have the same limitations of the not so quiet room when I'm listening to the headphone. and while I measure stuff I fail to hear all the time, I'm can't say I have any example of the opposite, where I can positively claim it's not all in my head, but it doesn't clearly manifest on a recording. I would need to know clearly what to look for, set up a blind test for it, find it, and have nothing on the measurements I'd make under the same circumstances. I have no knowledge of such event being clearly documented. has it ever happen? was it never later discovered that it was just a testing error or some not so blind test with a clever horse as test subject? I'm always willing to admit that I don't know much and that I'm wrong. it happens all the time and I have no issue with that. on the contrary when I notice it it makes me a better man the next day with one less wrong in my mind. I want to know the truth, not just to win on internet. but you have to give me something. just telling me to be open minded and to pretend that everything is always the same, that's really not doing me any good.
I have to insist about a dire need for some reference in the magnitudes we must consider relevant. and a name for the variables we're supposed to look for. without those, this discussion is a total waste of my time. everybody can be right and wrong, it's not even about moving the goalpost, nobody even bothered to put some on the field yet.

so please whoever knows something beyond having a lot of self confidence, please pretty please, come forward and give us something to look for in headphones. otherwise we can just rename this topic coca vs pepsi and we wouldn't lose anything to the relevance of the arguments.

 

[bigshot]

The fundamentals of recorded sound are frequency response, amplitude and distortion. Everything fits in the broadest definition of those three terms. We can measure all those things. If there is anything else to sound reproduction it is directionality, and that doesn't apply to two channel stereo recordings of music since channel separation is pretty much perfect.

 

[jagwap]

to me this would come down to 2 possibilities:
1/ there is a physical change in the drivers, but they don't end up affecting the outgoing signal in any significant way. so who cares?
2/ BS excuse. why wouldn't we be able to measure clearly audible change from the same source with the same mic over time? that's something I really don't get. if there is something I'm missing you guys need to explain it to me once and for all. characterizing a change could be real hard. identifying one variable as sole cause of a subjective impressions is often impossible. identifying a cause would probably require to observe the driver directly with lasers or other optical solution at really high speed and accuracy instead of looking at the sound. but if something is changing audibly over time, then it means the air pressure is changing in magnitude, else how do we hear it? and if a change is as I expect as simple as a change in the amplitudes of a signal over time, why should it be hard to record and confirm it, even with something as dumb as a null between 2 recordings at different time?
I just can't wrap my head around this. I have limitation because my rig is super cheap and my room not so quiet. but I have the same limitations of the not so quiet room when I'm listening to the headphone. and while I measure stuff I fail to hear all the time, I'm can't say I have any example of the opposite, where I can positively claim it's not all in my head, but it doesn't clearly manifest on a recording. I would need to know clearly what to look for, set up a blind test for it, find it, and have nothing on the measurements I'd make under the same circumstances. I have no knowledge of such event being clearly documented. has it ever happen? was it never later discovered that it was just a testing error or some not so blind test with a clever horse as test subject? I'm always willing to admit that I don't know much and that I'm wrong. it happens all the time and I have no issue with that. on the contrary when I notice it it makes me a better man the next day with one less wrong in my mind. I want to know the truth, not just to win on internet. but you have to give me something. just telling me to be open minded and to pretend that everything is always the same, that's really not doing me any good.
I have to insist about a dire need for some reference in the magnitudes we must consider relevant. and a name for the variables we're supposed to look for. without those, this discussion is a total waste of my time. everybody can be right and wrong, it's not even about moving the goalpost, nobody even bothered to put some on the field yet.

so please whoever knows something beyond having a lot of self confidence, please pretty please, come forward and give us something to look for in headphones. otherwise we can just rename this topic coca vs pepsi and we wouldn't lose anything to the relevance of the arguments.

Again, magnitude isn't everything. I think we need to focus elsewhere like phase and decay. This may show up in an impulse response, but as you say laser interferometry is more likely to show up any differences. Then any changes in modal breakup or surround behaviour would be available.

In the possible flawed innerfidelity data the impulse response does trend towards larger peaks, and the low frequencies appear to extend lower as burn in continues. The LF changes are small, but that can mean a significant change in the first pole, leading to significant phase shift in the audio band. This may align with the Fs changes etc of the woofer data?

 

[bigshot]

Phase is distortion. Decay is amplitude.

 

[jagwap]

The fundamentals of recorded sound are frequency response, amplitude and distortion. Everything fits in the broadest definition of those three terms. We can measure all those things. If there is anything else to sound reproduction it is directionality, and that doesn't apply to two channel stereo recordings of music since channel separation is pretty much perfect.

As I said, the macro effects. But there are others which are important. You've never used the trick of altering the LF phase to change the perceived pace of a peice of music?

 

[jagwap]

Phase is distortion. Decay is amplitude.

Yet it doesn't show up when you do a sweep of a speaker crossover.

Conventional measurements do not show these "distortions". You have to go look for them. THD+N does not show phase errors. Frequency response plots don't show decay or Q readily. If they did there would be a lot less badly ported loudspeakers out there.

 

[gregorio]

This is back to the circular arguement: SS: We cannot measure it so it is not there.

Yes, it is a circular argument but it's a circular argument to which there is no logical counter argument! If we cannot measure it, then we cannot record it so obviously "it is not there" in the recording and, just as obviously, if it is not there in the recording then how can we reproduce it?!

How about this: the measurement I linked to show a significant change on a woofer. Things like Fs changed by over 10%.

I can't find the measurement you linked to, which would be useful but even without that, your statement raises several points: 1. It's a measurement then? 2. Do you typically strap woofers to your head? 3. I'm sure it's possible to design a woofer driver, probably even a headphone driver and a microphone capsule which does require burn-in, the question is, why would you want to? It would play havoc with the design tolerances of the other elements/components of your product, havoc with your published specifications and potentially havoc with both your product's reliability and consumer satisfaction and for what benefit? I'm sure there are at least some incompetent designers out there and I'm equally sure there are examples of manufacturers who've used the excuse of burn-in to fob-off consumers unsatisfied with a poor design or poor quality control.

Conventional measurements do not show these "distortions".

Again, all we can record is frequency and amplitude and therefore all we can reproduce is frequency or amplitude. If any piece of equipment causes any distortion (change) to that frequency and/or amplitude then we can measure the difference between the input frequency/amplitude and the output frequency/amplitude. If we cannot measure a difference, then there isn't one!

G

 

[donunus]

After all the circular arguments I think we have all learned that in the end people believe what they want to believe and never trust anyone else with experience of opposing results. The thing is, there is no proof that can be demonstrated for this except to trust one's ears while having a benchmark to go back to to reduce placebo.

 

[jagwap]

Yes, it is a circular argument but it's a circular argument to which there is no logical counter argument! If we cannot measure it, then we cannot record it so obviously "it is not there" in the recording and, just as obviously, if it is not there in the recording then how can we reproduce it?!

We couldn't measure TIM distortion before the '70s, but it didn't mean it didn't exist. What makes you think audio is fully discovered? Have AES papers stopped being published. Also I am not stating we cannot measure it. I am suggesting we do not discover it with a simplistic view that frequency response tells us all we need to know.

I can't find the measurement you linked to, which would be useful but even without that, your statement raises several points: 1. It's a measurement then?

Look again, I'm not your Google. Yes it is a bunch of speaker Thiele-Small parameters that showed noticable change in the first 20 hours on a woofer.[/QUOTE]

2. Do you typically strap woofers to your head?

Yes, don't you? Ask a stupid question and wonder why my respect deminishes by the reply. It's a question if scale. If Castleofargh can argue this coherently, perhaps you can learn from him. I suspect you think you like learning more than you actually do.

3. I'm sure it's possible to design a woofer driver, probably even a headphone driver and a microphone capsule which does require burn-in, the question is, why would you want to? It would play havoc with the design tolerances of the other elements/components of your product, havoc with your published specifications and potentially havoc with both your product's reliability and consumer satisfaction and for what benefit? I'm sure there are at least some incompetent designers out there and I'm equally sure there are examples of manufacturers who've used the excuse of burn-in to fob-off consumers unsatisfied with a poor design or poor quality control.

Nobody wants burn-in. It appears to be laws of physics getting in the way of audio perfection again. Again a silly point aimed at annoying people.

Again, all we can record is frequency and amplitude and therefore all we can reproduce is frequency or amplitude. If any piece of equipment causes any distortion (change) to that frequency and/or amplitude then we can measure the difference between the input frequency/amplitude and the output frequency/amplitude. If we cannot measure a difference, then there isn't one!

G

That is not what is recorded. There is nothing in PCM that says frequency. We record a sequence of amplitudes, which together when kept in sequence and in time represent a signal. When they repeat with a regular interval that repetition causes frequencies to be heard. But you know this.

 

[gregorio]

[1] Nobody wants burn-in. It appears to be laws of physics getting in the way of audio perfection again.
[2] That is not what is recorded. There is nothing in PCM that says frequency. We record a sequence of amplitudes, which together when kept in sequence and in time represent a signal.

1. Great, so my measurements of my speakers and mics and the measurements of many other sound/music engineers have broken the laws of physics, thanks for that useful assessment!
2. And what signal is it that we are representing? Oh dear!

G

 

[jagwap]

1. Great, so my measurements of my speakers and mics and the measurements of many other sound/music engineers have broken the laws of physics, thanks for that useful assessment!
2. And what signal is it that we are representing? Oh dear!

G

Now it is clear you delibarately misunderstand what is written. To what purpose?

 

[castleofargh]

Again, magnitude isn't everything. I think we need to focus elsewhere like phase and decay. This may show up in an impulse response, but as you say laser interferometry is more likely to show up any differences. Then any changes in modal breakup or surround behaviour would be available.

In the possible flawed innerfidelity data the impulse response does trend towards larger peaks, and the low frequencies appear to extend lower as burn in continues. The LF changes are small, but that can mean a significant change in the first pole, leading to significant phase shift in the audio band. This may align with the Fs changes etc of the woofer data?

again, FR from a sweep or averaged noise are convenient but not necessary unless we know what we're looking for. impulse could perhaps do they're not the easiest stuff to capture properly and consistently being so short and all, but if you feel confident that driver change would clearly come out as something consistent and much bigger than artifacts from measurements, then I'm fine with that.
or we could just as well record a song twice. so long as we have amplitude over time, we have the sound. if audible change comes from the sound we should see it. even with a simple null of the 2 records for example. it would give us variations and a bunch of noise from various origins. if the variations from the driver are significant it will show there. we could also have people ABX it. so the group of "we can't measure everything in sound" can enjoy their listening test to confirm change. although I imagine a good portion of those with that argument will somehow also have something to say about controlled listening tests. but at least those are viable methods to me.
we have to start by capturing those changes in a way that has the driver in a fixed position as unequivocal cause for the change. which is why I'd like to forget all about that innerfidelity test. the low end boost over time is at least in part, a direct result of the pads on my 2 headphones. his test was cool, he was right to do it and it showed changes. just no clue if the driver caused most or any of them. in that respect it's not useful at all.

After all the circular arguments I think we have all learned that in the end people believe what they want to believe and never trust anyone else with experience of opposing results. The thing is, there is no proof that can be demonstrated for this except to trust one's ears while having a benchmark to go back to to reduce placebo.

well in this section taking subjective impressions from sighted test at face value is seen as a lapse in judgement. I'm sure some will reject an impression only because it doesn't align with their views, yet accept another when it does. boys will be boys. but rejecting impressions from uncontrolled tests is the proper decision, whatever those impressions are in favor of. when bigshot says he didn't hear a difference, that does nothing to disprove driver burn in. when you say you noticed a change, it does nothing to prove driver burn in.
I care about making that clear more than I care about finding a consistent model to characterize and predict burn in. audiophile are so used to have to manage only with subjective impressions, that they somehow started thinking they were evidence of objective events, and for some, the only evidence. which of course is a mistake. science doesn't demonstrate anything by going around and asking how people feel. we do that when we care about people's feeling, not about objective change in sound.

 

[gregorio]

Now it is clear you delibarately misunderstand what is written.

How have I misunderstood? Sure, we cannot design perfect transducers but I have measured numerous transducers when new and after many hours use and have not detected any change in the measurements. Are my measurements braking the laws of physics or is it possible to design a transducer which while not perfect, does not burn-in/change noticeably? And, if it's possible to design a transducer which doesn't burn-in and "nobody wants burn-in" then why would some other transducers burn-in?

Your answer to this appears to be, you cannot measure everything but how is that relevant? We don't need to measure everything, just the input signal and the output signal (or two output signals at different times). Then all we need to do is compare them for any difference, NOT measure everything!

G